• English 
  • Spanish 


Wind speed and power supply prediction in wind farms using statistical methods. Economic value from the electric power market perspective

Project type: National project
Program: Spanish R+D Program 2004-2007
Code: CGL2005-06966-C07-02/CLI
Period: October 2005 - October 2008
Status: Finished

This project is part of a coordinated project (9 subprojects) to undertake a study on meteorological prediction applied to wind resource management, focusing on short and middle-range wind power forecasting (up to 48-72 hours) as well as seasonal forecasting (up to 1 month) in complex terrain. Also, the economic value of each of the methodologies will be assessed from the point of view of the integration into the electrical dispatch system, considering practical aspects of its application as, for instance, the repercussion in the spot market shares. Thus, this project proposes an interesting application of Meteorology in an economic problem of national interest, regarding clean and sustainable technologies.

This problem can be approached either by considering directly wind farm power output, or by first obtaining wind forecasts in complex terrain and then deriving power output. In both cases, a model of the wind system is needed, including not only the average power curve, but also the dynamical characteristics of the wind farm (start/stop, hysteresis etc.) in order to get the intended forecasted variable. In the proposed project IN-VENTO, both approaches are considered, extending the current state of the knowledge and the actual know-how, with the aim of producing regional and local simulations of wind and wind power, applying linear and non-linear statistical models, downscaling methods, micro and mesoscale dynamical models and power curve models. Therefore, all methods provided by the available technology will be applied in an integrated way. Some of them have been already applied (mainly, linear statistical models) and other are new methods, more adequate to our problem. Also, an economic model will be developed based on electricity market rules, for comparing the value from the distinct methodologies in different time periods and sites.

As a start point, the available data bases on wind speed and direction and wind power will be analysed in detail for distinct sites. The probability distributions and variability will be studied, in order to improve the knowledge about climate variability in relation to wind speed and power; moreover, the most influent processes at a local scale will be identified. Simultaneously, high-resolution integrations of the Iberian Peninsula will de realized, producing a reanalysis with higher resolution then the actual ones. This new reanalysis (in the same period as ERA-40) will be appropriate for the application and validation of the methods developed in this project, including optimised parametrizations for wind, a surface model and data assimilation techniques.

The high-resolution integrations provide a first estimation for wind speed, but some methods will be applied on these integrations with the intention of obtaining more precise predictions: statistical methods (using also the available observations), microscale methods (considering local effects of orography, obstacles etc.) and dynamical models of the wind farm behaviour, of the wind turbines and power. The economic model to de developed will permit an estimation on the validity of each one of the different methodologies. Still, an automatic system for taking decisions will be developed, based on all forecasts and their respective economic values. This system will be oriented to solve interesting questions for the companies interested in this project (e.g., power prediction 6 hours ahead, etc.).

The final conclusion of this project will be an evaluation about the actual capacity of the meteorological prediction applied to the wind resource management and the integration into the new scenario (near future) of the national electrical system, considering the spot market (short-range prediction) and also giving estimations on future tendencies (seasonal prediction).

Contribution: Our group coordinates the sub-project dealing with statistical downscaling methods, represented in the upper-left panel of the figure below.